🚀 Better prompts. Smarter agents. More control. The latest Visual Studio Code release introduces the Chat Customizations Evaluation extension, designed to help analyze and refine your prompts, agents, instructions, and skills.

🔎 Also in this release: chat session insights, codebase search across workspaces & GitHub, dedicated skill context, and more. 🔗 Explore the weekly updates: Happy coding! 💙

Another bullshit feature. Already frustrated with copilot session limits and weekly limits. I still had 20% left on my premium request, but you put the weekly limit on it, and now it's resetting on day one.

it doesnt support agents.md :') only instructions and prompts under .github

This will save so many broken agent runs. Installing now.

Stoooop! Just fucking STOP! Please keep an IDE an IDE, and focus on the IDE features. Either make an entity separate from the VSCode (not legally, just practically) stuff, or treat everything that is not IDE related (all AI stuff) as the, at least, second most important thing.

I want to start off by saying, love VS Code, I use it everyday for all my projects. However, I had to downgrade to 1.117 after upgrading an constantly had my extensions host crash. I ran the Extension Bisect and after starting the application without any installed extensions, the extensions host continues to crash. Please fix 🙏

We all moved to skills and AGENTS.md files ...what now ???

Extension logo looks like @Anydo ‘ s logo

It could work with AGENTS.md and SKILLS.md

If CU matters, develop software that optimizes cloud CU use, not 'waste' it in different ways. For example, where is the embedded general phi model and utilities that uses it?

fr this is the update i've been waiting for ngl

Hello,👋 Code Companion Mobile, which I’ve been developing for quite some time, is now available on the App Store! It lets you code from your phone as if you were using the IDE, using your Visual Studio Code Copilot session.

does this consume usage credits?

Couldn’t think of a worse name for this extension.

this is interesting

Carve flowers on shit

yeah rocked the new update :)

I don't prompt. Neither have I use for agents. What improvements does VS Code hold for non-ai developers?

ngl analyzing prompts sounds useful but also feels kinda meta. like i'd rather just see if the agent actually does the thing than get a breakdown of why it might not

Chat Customizations Evaluations auditing .prompt.md files in real time? Prompt intelligence just leveled up. No more broken agents.This is the cleanest upgrade yet.

You can already do that with agents and skills there is no need to slop bloatware. I know microslop loves complexity but you better drop github copilot prices and return premium models to people who trusted you and bought a year long plan.

Prompt eval is the gap I keep hitting — when I tweak a Claude skill or agent prompt, it's manual case-by-case checking, no clean regression tracking. Does this extension diff against held-out chat history, or do you specify expected outputs upfront?

Neat! Trying this tomorrow 🔥

not even the official page uses the dogshit new dark theme lmao